Towards Generalizable Deepfake Detection with Locality-aware AutoEncoder

Du, Mengnan; Pentyala, Shiva; Li, Yuening; Hu, Xia

doi:10.1145/3340531.3411892

Cited by 85 publications

(63 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Shortcut Learning Phenomena. Recently, the community has revealed the shortcut learning phenomenon for different kinds of language and vision tasks, such as NLI (Niven and Kao, 2019), question answering (Mudrakarta et al, 2018), reading comprehension (Si et al, 2019), VQA (Agrawal et al, 2018;Manjunatha et al, 2019), and deepfake detection (Du et al, 2020). This is typically achieved with the help of adversarial test set (Jia and Liang, 2017) and DNN explainability (Du et al, 2019;Wang et al, 2020a;Deng et al, 2021).…”

Section: Related Workmentioning

confidence: 99%

Towards Interpreting and Mitigating Shortcut Learning Behavior of NLU models

Du¹,

Manjunatha²,

Jain³

et al. 2021

Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Langua

Self Cite

View full text Add to dashboard Cite

Recent studies indicate that NLU models are prone to rely on shortcut features for prediction, without achieving true language understanding. As a result, these models fail to generalize to real-world out-of-distribution data. In this work, we show that the words in the NLU training set can be modeled as a longtailed distribution. There are two findings: 1) NLU models have strong preference for features located at the head of the long-tailed distribution, and 2) Shortcut features are picked up during very early few iterations of the model training. These two observations are further employed to formulate a measurement which can quantify the shortcut degree of each training sample. Based on this shortcut measurement, we propose a shortcut mitigation framework LTGR, to suppress the model from making overconfident predictions for samples with large shortcut degree. Experimental results on three NLU benchmarks demonstrate that our long-tailed distribution explanation accurately reflects the shortcut learning behavior of NLU models. Experimental analysis further indicates that LTGR can improve the generalization accuracy on OOD data, while preserving the accuracy on in-distribution data. Input x Teacher model Softmax Softmax Ground truth y Smoothed Softmax Distill loss Student loss Overconfident prediction Long-tailed distribution Example of model paying high attention to features on the head (a) long-tailed observation (b) Mitigation framework Shortcut degree Student model Head Long tail Shortcut degree Data statistics Model behavior

show abstract

Section: Related Workmentioning

confidence: 99%

Towards Interpreting and Mitigating Shortcut Learning Behavior of NLU models

Du¹,

Manjunatha²,

Jain³

et al. 2021

Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Langua

Self Cite

View full text Add to dashboard Cite

show abstract

“…Although researchers in the community have investigated the DeepFake detection problem from various perspectives, only minimal effort has been devoted to investigating DeepFakes from a fine-grained visual classification point of view, especially using attentionbased techniques. The most similar works to ADD are [26,57] methods. In line with [26,57], our proposed method looks at the DeepFake detection problem as a fine-grained visual classification task while utilizing attention-based data augmentation techniques.…”

Section: Related Workmentioning

confidence: 99%

“…As DeepFakes became super-realistic and more pervasive, ascertaining a digital video's trustworthiness and deciding on its authenticity becomes a more demanding yet challenging task. The fact that DeepFakes are created exploiting an AI algorithm rather than a camera capturing real events implies that they can still be detected using advanced deep learning networks [26]. Recently, multiple research works have focused on presenting a comprehensive understanding of the state-of-the-art methods and comparative analysis of DeepFakes [27][28][29].…”

Section: Introductionmentioning

confidence: 99%

ADD: Attention-Based DeepFake Detection Approach

Khormali

Yuan

2021

BDCC

View full text Add to dashboard Cite

Recent advancements of Generative Adversarial Networks (GANs) pose emerging yet serious privacy risks threatening digital media’s integrity and trustworthiness, specifically digital video, through synthesizing hyper-realistic images and videos, i.e., DeepFakes. The need for ascertaining the trustworthiness of digital media calls for automatic yet accurate DeepFake detection algorithms. This paper presents an attention-based DeepFake detection (ADD) method that exploits the fine-grained and spatial locality attributes of artificially synthesized videos for enhanced detection. ADD framework is composed of two main components including face close-up and face shut-off data augmentation methods and is applicable to any classifier based on convolutional neural network architecture. ADD first locates potentially manipulated areas of the input image to extract representative features. Second, the detection model is forced to pay more attention to these forgery regions in the decision-making process through a particular focus on interpreting the sample in the learning phase. ADD’s performance is evaluated against two challenging datasets of DeepFake forensics, i.e., Celeb-DF (V2) and WildDeepFake. We demonstrated the generalization of ADD by evaluating four popular classifiers, namely VGGNet, ResNet, Xception, and MobileNet. The obtained results demonstrate that ADD can boost the detection performance of all four baseline classifiers significantly on both benchmark datasets. Particularly, ADD with ResNet backbone detects DeepFakes with more than 98.3% on Celeb-DF (V2), outperforming state-of-the-art DeepFake detection methods.

show abstract

“…Du et al. [40] constructed a novel detection method to improve the generalization accuracy by making predictions relying on correct forgery evidence. Nguyen et al.…”

Section: Related Workmentioning

confidence: 99%

Deep forgery discriminator via image degradation analysis

Zhang

et al. 2021

IET Image Processing

View full text Add to dashboard Cite

Generative adversarial network-based deep generative model is widely applied in creating hyper-realistic face-swapping images and videos. However, its malicious use has posed a great threat to online contents, thus making detecting the authenticity of images and videos a tricky task. Most of the existing detection methods are only suitable for one type of forgery and only work for low-quality tampered images, restricting their applications. This paper concerns the construction of a novel discriminator with better comprehensive capabilities. Through analysis of the visual characteristics of manipulated images from the perspective of image quality, it is revealed that the synthesized face does have different degrees of quality degradation compared to the source content. Therefore, several kinds of image quality-related handicraft features are extracted, including texture, sharpness, frequency domain features, and deep features, to unveil the inconsistent information and modification traces in the fake faces. In this way, a 1065-dimensional vector of each image is obtained through multi-feature fusion, and it is then fed into RF to train a targeted binary classification detector. Extensive experiments have shown that the proposed scheme is superior to the previous methods in recognition accuracy on multiple manipulation databases including the Celeb-DF database with better visual quality.

show abstract

Towards Generalizable Deepfake Detection with Locality-aware AutoEncoder

Cited by 85 publications

References 20 publications

Towards Interpreting and Mitigating Shortcut Learning Behavior of NLU models

Towards Interpreting and Mitigating Shortcut Learning Behavior of NLU models

ADD: Attention-Based DeepFake Detection Approach

Deep forgery discriminator via image degradation analysis

Contact Info

Product

Resources

About